Extracting value from large amounts of data {and making human sense of it is one of the primary challenge of data science

Introduction to Data Science

1.Find the data

2.Extract and acquire the data

3.Clean and transform the data

4.Understand the relationships in the data
and build a model

5.Mine for additional data

6.Evaluate and refine the model

7.Communicate the results

Applied Data Science at Microsoft

Cortana {contextual understanding}

Bing {ranking, recommendations}

Windows 10 {Insider telemetry}

Windows Hello {next gen security}

Skype Translator {voice input}

HoloLens {object recognition}

Xbox {gamer matchmaking}

Microsoft Research {innovation}

Microsoft Cognitive Toolkit

Speed & scalability {while maintaining accuracy}

Commercial-grade AI {used in Cortana & Bing}

Familiar {use Python on Linux or Windows}

C#, R, Spark, Docker, and other open source coming soon

So What is CNTK

CNTK expresses (nearly) arbitrary neural networks by composing simple building blocks into complex computational networks, supporting relevant Network types and applications.

CNTK is production-ready: State-of-the-art accuracy, efficient, and scales to multi-GPU/multi-server.

Lego-like composability allows CNTK to support a wide range of networks, e.g.

• feed-forward DNN
• RNN, LSTM, GRU
• convolution
• DSSM

sequence-to-sequence

CNTK is ideal for a range of applications including

1. speech
2. vision
3. text
4. and combinations
5. Large data sets is the key here; performant

Microsoft’s open-source deep-learning toolkit

• Ease of use: what, not how
• Fast
• Flexible
• First class on Linux and Windows
• OpenSource

Getting Started with CNTK

https://notebooks.azure.com/n/1zbIwzaANic/notebooks/CNTK_101_LogisticRegression.ipynb

1.Configure reader, network, learner

2.Train & evaluate (multi-phase for layer building)

3.Deploy offline from Python

Binaries & Tutorials

https://github.com/Microsoft/CNTK/wiki

Deep-learning VM Toolkit https://azure.microsoft.com/en-us/marketplace/partners/microsoft-ads/dsvm-deep-learningtoolkit/

Deep-dive lecture & labs

https://www.youtube.com/watch?v=pl-kbFJ1KzU

Team Q&A

http://stackoverflow.com/questions/tagged/cntk

Escience Whats New in CNTK 2.0 https://esciencegroup.com/2016/11/10/cntk-revisited-a-new-deep-learning-toolkit-release-from-microsoft/ 

CNTK with GPU Example Using Azure NV12 Series Servers with 12 Cores 128GB Ram and 2 x M60 NVIDIA GPU

Getting Started with Microsoft Cognitive Toolkit  Webinar